The Retention & LTV Guide

Four in five customers never come back

Start with the number that should frame every retention conversation. Across 156,110 DTC customers, the repeat purchase rate was 18.8%, which means 81.2% bought exactly once and never returned (BS&Co, Repeat Purchase Rate Benchmarks, 156K customers, US data). That is materially lower than the 25-30% figure most operators quote, and the gap itself is the first lesson: blended averages hide where most stores actually live, which is the high teens and low twenties.

The Australian market makes the same point from the other direction. Australia Post's Inside Australian Online Shopping reporting found 82% of Australian households shopped online in 2025 (9.8 million households, a record), and that the average order value is now drifting down towards roughly A$96 while purchase frequency rises by about four orders a year (Australia Post eCommerce Report 2025-2026). A market where customers buy more often but spend less per order is a market where retention, not basket size, decides who wins. The repeat relationship is the asset.

The averages mislead because they are pulled up by a minority of replenishment-driven and subscription brands. Shopify's own merchant data clusters the typical repeat rate near 27-28% (Mobiloud, BLOY syntheses of Shopify data), but the median single store sits lower. If you benchmark yourself against the 28% headline, you will conclude your retention is broken when it is in fact ordinary, or worse, you will conclude it is fine when it is quietly bleeding.

Why retention is worth fighting for is not a matter of opinion. Bain & Company's classic finding is that increasing customer retention by 5% can raise profit by 25% to 95%, and that acquiring a new customer costs five to twenty-five times more than retaining an existing one (Bain & Company, Retaining customers is the real challenge). The probability of selling to an existing customer is roughly 60-70% versus 5-20% for a new prospect, and existing customers spend about 31% more on average. Retention is not a loyalty nicety. It is the cheapest growth a business owns.

The second order is everything

If the first purchase is a stranger taking a chance, the second is the moment a buyer becomes a customer. The data is unusually clean here. The foundational RJMetrics benchmark of 176 retailers and 18 million customers found that after a first order a customer has only a 32% chance of placing a second, but once they place that second order the chance of a third climbs to 54% (RJMetrics, Ecommerce Buyer Behavior, 2015). Each completed order raises the conditional probability of the next. Retention compounds, and the compounding starts at order two.

This reframes the whole acquisition argument. The marginal cost of converting a one-time buyer into a two-time buyer is almost always lower than the cost of acquiring a fresh first-time buyer, because you already own the relationship, the contact details, and the consent to message them. The second order is the highest-return dollar in the funnel, and most brands spend nothing deliberate on it.

The shape of that flatten matters more than the starting height. A cohort that flattens at 31% instead of 14% has roughly 2.2x the long-run customer base from the same acquisition spend. You did not acquire more customers. You kept more of the ones you already paid for, and that multiple flows straight through to lifetime value.

Reading retention curves: declining, flattening, smiling

A retention curve plots the share of a cohort still active against time since first purchase. There are three canonical shapes, and which one you have is the single most important diagnosis in the business (Userpilot, Churnkey, Shopify cohort-analysis guides).

• Declining. The curve keeps falling and never flattens. This is the signature of no product-market fit: you are renting customers, not earning them. No amount of lifecycle marketing fixes a declining curve, because the problem is the product or the fit, not the messaging.
• Flattening. A steep early drop, then a stable floor. This is the most common healthy DTC pattern. The height of the floor is your core loyal-customer rate, the share who will keep buying more or less indefinitely. The earlier and higher a cohort flattens, the healthier the business.
• Smiling. The curve drops, stabilises, then rises as dormant customers reactivate or accounts expand. This is the strongest signal there is, and it usually means a working win-back programme, a subscription, or genuine expansion revenue.

Fitting the curve: the retention power law

Practitioners fit retention curves with a power law, not an exponential, because real retention has a heavy tail: the customers who survive the first few months churn far more slowly thereafter. The standard form is below, fit by ordinary least squares on the log-log transform (Shopify cohort guide; miniwebtool power-law cohort calculator).

With the fitted model you can do two things you could not do with a raw table. First, you can extrapolate the steady state: the loyal floor is where the curve stops falling materially, which the model places near a 19% retained share at month 24 for these demonstrative points. Second, you can compare cohorts on one number. A cohort with k = 0.35 is healthier than one with k = 0.55 even if their month-3 numbers look similar, because the lower exponent means a flatter, more valuable tail. The exponent k, not the headline repeat rate, is the metric to manage.

Building the cohort table that the curve comes from

The fit is only as honest as the cohort table beneath it, and there are three decisions that quietly change the answer. First, what counts as "active." For a transactional ecommerce business, active means "placed an order in the period," so the curve is a repeat-purchase curve. For a subscription business, active means "subscription not cancelled," a different and usually higher curve. Mixing the two definitions in one chart produces a number that means nothing.

Second, the cohort grain. Group customers by the month of first purchase, then track each monthly cohort forward. This isolates the customers acquired in, say, a heavy discount month from those acquired organically, and it is common for discounted cohorts to show a much steeper k. A blended all-customers curve hides that, and hiding it is how brands keep paying to acquire customers who were never going to come back.

Third, recency and the open tail. The most recent cohorts have not had time to mature, so their later-month points are missing rather than zero. Fit only on cohorts old enough to have observed the months you are using, or the regression will read the missing tail as churn and overstate k. This is the cohort equivalent of survivorship bias, and it is the most common reason a retention chart looks worse than the business actually is.

Once the table is clean, layering an RFM cut (recency, frequency, monetary value) on top turns the single curve into a segment map. The customers in the top frequency-and-monetary cell are the loyal floor made concrete; the recent-but-low-frequency cell is where the second-order programme should aim; the high-value-but-lapsing cell is the win-back priority list. RFM is textbook, decades old, and still the cheapest way to decide who gets which message.

Retention is not one number

A single company-wide repeat rate is almost useless, because retention is driven by the natural purchase cycle of the category. Consumables get re-bought because they run out; furniture does not. Comparing a supplement brand's repeat rate to a sofa brand's is a category error, literally. The dispersion is wide and predictable (Finsi.ai, Mageloyalty 2026 retention-by-industry data).

LTV done honestly, across service and ecommerce

Lifetime value is the most abused number in marketing because the horizon is a free parameter. Stretch the lifespan assumption and LTV inflates; the LTV:CAC ratio that looks like 5:1 on a five-year horizon is 2:1 on a two-year horizon. The standard multiplicative definition is honest only if every input is grounded.

The real test of unit economics is not the LTV:CAC ratio in isolation. A 3:1 ratio is the widely cited minimum, 4-5:1 is strong, but the ratio is meaningless without payback context and is trivially gamed by stretching the LTV horizon (Eightx, Airtree Ventures). The honest test is two conditions at once: LTV:CAC of 3:1 or better AND CAC payback under 12 months. A business can hit 5:1 on paper and still run out of cash if payback is 18 months, because the spend is upfront and the returns arrive in lumps.

For blended businesses that sell both a service and products, which describes a large share of Blufire's client base, LTV has to be assembled across both engines. A roofing or HVAC customer has a first-job value, then a repeat-and-referral tail; an ecommerce arm adds a separate CLV stream. Decomposing the blend is the only way to see which engine actually carries the relationship.

Email and SMS are not retention. They are acquisition you already paid for.

The common framing treats email and SMS as "retention channels," a soft category that sits in the corner of the budget. That framing costs money. Owned channels are the mechanism by which a retained customer's next order actually happens, and that next order is a sale you do not pay a platform to win. Revenue driven through owned channels is, in effect, acquisition cost you have already incurred. It is the cheapest demand in the business.

The performance gap between automated flows and one-off campaigns is enormous and consistent. In Klaviyo's benchmark across its global install base, email flows generated nearly 41% of total email revenue from just 5.3% of sends, with revenue per recipient roughly 18 times higher than campaigns (Klaviyo, 2024-2026 Email Marketing & SMS Benchmark Report; global, USD). SMS shows the same pattern: SMS flows drove 45.2% of total SMS revenue from 7.6% of sends (Klaviyo SMS benchmarks). The lesson is blunt: the money is in triggered, intent-based automation, not in batch-and-blast.

One more figure reframes the whole "retention vs acquisition" debate. In Klaviyo's data, nearly 48% of flow-driven email revenue came from new buyers, versus just 16% from campaigns. Welcome series, browse-abandonment, and cart-recovery flows are doing first-purchase conversion work, not just repeat work. The channel you filed under "retention" is quietly one of your best acquisition assets.

For sizing the prize, the headline channel-ROI figure is the one most operators have heard and few have sourced: email marketing returns roughly US$36 for every US$1 spent, higher than any other channel (Litmus / Data & Marketing Association ROI research, US data; the ratio is what matters and it is currency-neutral). Treat that as an order-of-magnitude ceiling for a well-run programme rather than a guarantee, but the direction is not in doubt: owned-channel revenue is the highest-margin demand a brand can generate.

The 30/60/90-day win-back window

Lifecycle timing is not a matter of taste. It is set by the empirical distribution of time-to-second-purchase, which is sharp. From the same 156K-customer dataset, the cumulative share of repeat buyers making their second purchase is 50.3% within 30 days, 76.4% within 90 days, and 96.3% within a year (BS&Co; corroborated by Klaviyo community data). Only 3.7% of second orders happen after a year.

That distribution has a long right tail, which means the mean time-to-second-order (50-100+ days) is misleading; the median sits at 15-35 days. Plan your post-purchase flows on the median, not the average, or you will send your win-back too late for the customers who were going to convert anyway and have nothing left for the ones who needed the nudge.

The practical sequence falls out of the table. The 30-day window is the peak, so the second-order nudge and the first replenishment primer belong there. The 60-90 day window is where a genuine win-back earns its keep, because the easy converters have already bought and the remaining buyers need a reason. Past 90 days you are in reactivation territory, and past a year the customer is, statistically, gone. Win-back timing is not a creative decision. It is a reading of where the mass of the distribution sits.

There is a parallel lesson for subscription and service models. Average monthly churn across 1,200+ subscription sites is 3.27%, split into voluntary (2.41%) and involuntary (0.86%) churn (Recurly Research, 2023 study). The involuntary slice, failed-payment churn, is the most recoverable revenue in the business: a dunning flow that retries cards and prompts updates routinely recovers a meaningful share of what would otherwise be silent attrition. It is the closest thing to free retention there is.

Subscription and recurring revenue: retention becomes the growth engine

When revenue recurs, retention stops being a marketing metric and becomes the growth model itself. The benchmark to know is net revenue retention (NRR), which measures how a cohort's revenue grows or shrinks over a year from expansion, contraction, and churn combined. Across private software companies the median NRR is roughly 101% and median gross revenue retention roughly 90%, with the top quartile above 92% on GRR (KeyBanc Capital Markets / Sapphire Ventures 2024 SaaS survey; SaaS Capital 2025 retention benchmarks). Best-in-class NRR runs 110-120%, and elite firms exceed 120%.

The reason operators obsess over NRR is its link to growth. Companies with NRR at or above 110% grow materially faster than the population median, and firms above 100% NRR grow roughly 1.8 times faster than lower-NRR peers (SaaS Capital). NRR above 100% means the existing base alone grows revenue with zero new logos, which is the cleanest expression of Profit Velocity there is: durable margin compounding off effort already spent.

The split between voluntary and involuntary churn is where most recovery hides. Involuntary churn, the 0.86% of customers lost to expired or failed cards, is not a satisfaction problem and does not need a discount to fix. It needs a payment-retry sequence, card-update prompts, and pre-dunning notices before the renewal date. Recurly reports that 54.5% of its customers saw churn decrease year on year in the 2024 analysis, and a disciplined dunning flow is consistently one of the largest single contributors. Recovering involuntary churn is the rare retention win that costs nothing in margin to capture.

Where retention meets Profit Velocity

Every thread in this guide ties back to one metric. Blufire's north-star is Profit Velocity: the rate at which a business converts marketing and sales effort into durable contribution margin. It rises when lifetime value grows and churn falls, because the numerator compounds and persists, and it rises when the cost and time to convert shrink.

What to do Monday

• Fit your curve. Pull cohort retention by month, run the log-log OLS, and read off k. That one exponent tells you whether you are flattening, declining, or smiling.
• Benchmark by category, not the market. Hold your repeat rate against your category band, not the 28% headline average.
• Move budget into flows. If campaigns dominate your lifecycle effort, you are working the low-leverage half. Build cart-recovery, welcome, browse-abandonment, and replenishment flows first.
• Time win-back to the median. Peak second-order window is 30 days; genuine win-back earns its keep at 60-90; reactivate past 90; concede past a year.
• Recover involuntary churn. A dunning flow is the cheapest retention dollar you can spend.
• Test LTV:CAC against payback. Never read the ratio without the payback period beside it.

Sources

BS&Co, Repeat Purchase Rate Benchmarks (156,110 DTC customers, US data). RJMetrics, Ecommerce Buyer Behavior benchmark, 2015 (176 retailers, 18M customers; 32% chance of a second order, 54% of a third; foundational study, US data). Australia Post, eCommerce Report / Inside Australian Online Shopping 2025-2026 (82% of AU households shopping online; AU AOV near A$96 and falling; frequency rising). Bain & Company, Retaining Customers Is the Real Challenge (Reichheld). Klaviyo, Email Marketing & SMS Benchmark Report (global install base, USD). Litmus / Data & Marketing Association, The ROI of Email Marketing (US$36 per US$1). Recurly Research, Customer Churn Benchmarks (1,200+ subscription sites). Finsi.ai, Repeat Purchase Rate and Ecommerce Retention Benchmarks (US data). Mageloyalty, Ecommerce Retention Benchmarks by Industry 2026. Salesforce AU and Pattern AU, Australian AOV benchmarks (A$150-A$161). Eightx and Airtree Ventures, LTV:CAC and CAC-payback guidance. KeyBanc Capital Markets / Sapphire Ventures and SaaS Capital, SaaS NRR/GRR benchmarks. Userpilot, Churnkey, and Shopify, retention-curve and cohort-analysis methodology.